Related Paper Analyzer

Purpose

Analyze user-provided papers and reports before final method selection.

This skill checks whether the user has placed original paper files under workspace/papers/, reads those originals, and extracts transferable method cues, assumptions, variables, validation ideas, and limitations for the current subquestions.

This skill does not fabricate references, browse for new papers, write the final method card, or copy a published model blindly.

When to use

Use this skill:

• After problem-parser and problem-classifier have produced validated artifacts.
• Before method-selector.
• When the team wants to ground method selection in user-supplied literature rather than guessing from model names.
• When original papers, reports, or extracted paper text are available under workspace/papers/.

Preconditions

The following should already exist or be provided:

• A validated problem parse.
• A validated problem classification artifact.
• At least one original paper file under workspace/papers/.

If no paper originals are present under workspace/papers/, stop and remind the user to place them there before analysis begins.

Inputs

Use or request:

• workspace/problem/problem-parser/problem_parse.json, if available.
• workspace/problem/problem-classifier/problem_classification.json, if available.
• Original paper files under workspace/papers/.
• User notes about which papers matter most, if available.

Workflow

1. Check workspace/papers/ first.

   • Look for user-supplied paper originals or extracted paper text.
   • Accept formats such as .pdf, .docx, .md, .txt, or clearly named extracted text files.
   • Ignore workspace/papers/related_paper_analysis.md if it already exists.

2. Stop early if the paper folder is empty.

   • If no paper originals are found, tell the user to place the paper originals under workspace/papers/.
   • Do not fabricate literature summaries from memory.

3. Inventory the available papers.

   • Record file name, apparent title if recoverable, likely language, and likely relevance.
   • Mark unreadable or weakly relevant files explicitly.

4. Read papers for modeling signals.

   • Extract task type, assumptions, variables, data requirements, method families, outputs, validation patterns, and limitations.
   • Focus on transferable ideas rather than copying a complete published pipeline.

5. Map literature cues to the current problem.

   • Link relevant papers to specific subquestions such as Q1, Q2, or Q3.
   • Distinguish useful method cues from paper-specific tricks that should not be copied directly.

6. Summarize risks and reuse boundaries.

   • Flag assumptions that may not fit the current contest problem.
   • Flag data requirements that the current workspace may not satisfy.
   • Flag methods that look impressive in papers but may be infeasible under contest constraints.

7. Write the literature analysis report.

   • Save the report as workspace/papers/related_paper_analysis.md.
   • Keep it concise, traceable, and ready for method-selector.

8. Hand off to method-selector.

   • Pass forward the paper inventory, method cues, cautions, and unresolved evidence gaps.

Outputs

Produce one Markdown literature-analysis artifact:

• workspace/papers/related_paper_analysis.md

The report should include:

• reviewed paper inventory
• per-paper summary
• subquestion-to-paper mapping
• transferable method cues
• assumptions or limitations worth carrying forward
• methods that should not be copied blindly
• missing evidence or unresolved questions
• recommended next skill

Output format

Write a Markdown report at workspace/papers/related_paper_analysis.md.

Suggested sections:

• Reviewed papers
• Transferable method cues by subquestion
• Useful assumptions, variables, and validation ideas
• Risks of direct reuse
• Missing evidence
• Recommended next skill